centralized error logging

[Dieterbe]

for tsdb benchmarks especially, but also in general, we need to be able to easily and reliably determine if there were any errors across the entire stack and what they were.
for this i wrote a little script that builds on the idea that all our containers log to /var/log/raintank, and that directory is mapped to the logs directory.
effedbc
it covers

dieter@dieter-m6800 raintank-docker ls -1 logs/
collector.log
create-influxdb-datasource.log
grafana-dev.log
graphite-api.log
graphite-app.log
graphite-flask.log
graphite-watcher.log
nsq_metrics_to_elasticsearch.log
nsq_metrics_to_kairos.log
nsq_probe_events_to_elasticsearch.log
prepare-nsqd.log

notably still missing (basically the list of non-raintank containers):

• cassandra
• kairosdb
• elasticsearch
• nsqd
• mysql
• redis

@woodsaj could you help with exposing errors in any of those? and doing a quick sanity check to confirm that my approach makes sense